# Segment Anything

**Authors:** Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollár, Ross Girshick  
**arXiv:** [2304.02643](https://arxiv.org/abs/2304.02643)  
**Date:** 2023-04-05  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2304.02643v1/

## Key points

- SA project introduces a new task and model for image segmentation
- Developed an efficient model in a data collection loop to create the largest segmentation dataset to date
- Dataset contains over 1 billion masks on 11 million licensed and privacy-respecting images
- Model is promptable, allowing it to transfer zero-shot to new image distributions and tasks
- Capabilities of the model were evaluated on numerous tasks and found impressive zero-shot performance, often competitive with or even superior to prior fully supervised results
- The team released the Segment Anything Model (SAM) and corresponding dataset (SA-1B) of 1 billion masks and 11 million images at https://segment-anything.com
- Authors of the project include Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollár and Ross Girshick

## Summary

The Segment Anything (SA) project introduces a new task and model for image segmentation. The team developed an efficient model in a data collection loop to create the largest segmentation dataset to date, containing over 1 billion masks on 11 million licensed and privacy-respecting images. This model is designed and trained to be promptable, allowing it to transfer zero-shot to new image distributions and tasks. The capabilities of this model were evaluated on numerous tasks and found that its zero-shot performance is impressive, often competitive with or even superior to prior fully supervised results. To promote research into foundation models for computer vision, the team released the Segment Anything Model (SAM) and corresponding dataset (SA-1B) of 1 billion masks and 11 million images at https://segment-anything.com. The authors of the project include Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollár and Ross Girshick.